Thinking Machines unveils Inkling, an open-weight audio-native LLM

Thinking Machines, a lab founded by former OpenAI CTO Mira Murati, has released Inkling, an open-weight model with native audio processing capabilities. This 975 billion parameter model, with 41 billion active parameters, can process text, images, and audio directly without external encoders, a feature previously exclusive to closed labs. While Inkling excels at understanding audio, transcribing speech, and processing voice instructions, it currently only responds in text, not synthesized speech. The model's performance on benchmarks like AudioMC shows a significant leap over open-weight competitors, though closed models still lead in overall capabilities. AI

IMPACT This release could accelerate the development of multimodal AI applications by providing an open-weight model with native audio understanding, though its lack of audio output limits immediate voice interaction use cases.

RANK_REASON Open-weight model release from a notable lab (ex-OpenAI CTO) with novel native audio processing capabilities. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on dev.to — LLM tag →

• ChatGPT
• DeepSeek V3
• Gemini
• Gemini 3.1 Pro
• GPT-4
• GPT-4o
• Hugging Face
• Inkling
• Mira Murati
• OpenAI
• Qwen3-Omni
• Thinking Machines

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →